Algorithmic basis of reward-based decisions in dynamic environment In probabilistic reward foraging task (Fig.1A) animals’ choices tend to reflect the previous rewards, in a way that biases animals’ future choices to options from which they recently received the rewards. In addition to the reward history effects, curiously animals choices also depend on the previous choices, phenomenon that is not easily explained by the current computational models (Fig.1B). We are formulating normative accounts of why choice history effects should persist in animals. The insight into this phenomenon may give
